Feature [feature-requests:#1458] Move code into member functions, use

string_view and constexpr.

1 files changed, 76 insertions, 106 deletions

diff --git a/src/CaseConvert.cxx b/src/CaseConvert.cxx
index 5beffbcc2..601432f14 100644
--- a/src/CaseConvert.cxx
+++ b/src/CaseConvert.cxx
@@ -31,7 +31,7 @@ namespace {
 // Another pattern (pitch==2) is where each lower case letter is preceded by
 // the upper case form. These are also grouped into ranges.
 
-int symmetricCaseConversionRanges[] = {
+constexpr int symmetricCaseConversionRanges[] = {
 //lower, upper, range length, range pitch
 //++Autogenerated -- start of section automatically generated
 //**\(\*\n\)
@@ -94,7 +94,7 @@ int symmetricCaseConversionRanges[] = {
 // Code points that are symmetric but don't fit into a range of similar characters
 // are listed here.
 
-int symmetricCaseConversions[] = {
+constexpr int symmetricCaseConversions[] = {
 //lower, upper
 //++Autogenerated -- start of section automatically generated
 //**1 \(\*\n\)
@@ -262,7 +262,7 @@ int symmetricCaseConversions[] = {
 // folding is different to lowering, or (as appropriate) upper(lower(x)) != x or
 // lower(upper(x)) != x.
 
-const char *complexCaseConversions =
+constexpr std::string_view complexCaseConversions =
 // Original | Folded | Upper | Lower |
 //++Autogenerated -- start of section automatically generated
 //**2 \(\*\n\)
@@ -577,22 +577,20 @@ const char *complexCaseConversions =
 //--Autogenerated -- end of section automatically generated
 ;
 
+// Maximum length of a case conversion result is 6 bytes in UTF-8
+constexpr size_t maxConversionLength = 6;
+
 class CaseConverter : public ICaseConverter {
-	// Maximum length of a case conversion result is 6 bytes in UTF-8
-	enum { maxConversionLength=6 };
 	struct ConversionString {
-		char conversion[maxConversionLength+1];
-		ConversionString() noexcept : conversion{} {
-		}
+		char conversion[maxConversionLength+1]{};
 	};
 	// Conversions are initially store in a vector of structs but then decomposed into
 	// parallel arrays as that is about 10% faster to search.
 	struct CharacterConversion {
-		int character;
+		int character = 0;
 		ConversionString conversion;
-		CharacterConversion() noexcept : character(0) {
-			// Empty case: NUL -> "".
-		}
+		// Empty case: NUL -> "".
+		CharacterConversion() noexcept = default;
 		CharacterConversion(int character_, std::string_view conversion_) noexcept : character(character_) {
 			assert(conversion_.length() <= maxConversionLength);
 			try {
@@ -607,6 +605,7 @@ class CaseConverter : public ICaseConverter {
 			return character < other.character;
 		}
 	};
+	CaseConversion conversion;
 	typedef std::vector<CharacterConversion> CharacterToConversion;
 	CharacterToConversion characterToConversion;
 	// The parallel arrays
@@ -614,7 +613,8 @@ class CaseConverter : public ICaseConverter {
 	std::vector<ConversionString> conversions;
 
 public:
-	CaseConverter() = default;
+	explicit CaseConverter(CaseConversion conversion_) : conversion(conversion_) {
+	};
 	// Deleted so CaseConverter objects can not be copied.
 	CaseConverter(const CaseConverter &) = delete;
 	CaseConverter(CaseConverter &&) = delete;
@@ -624,8 +624,8 @@ public:
 	bool Initialised() const noexcept {
 		return !characters.empty();
 	}
-	void Add(int character, const char *conversion) {
-		characterToConversion.emplace_back(character, conversion);
+	void Add(int character, std::string_view conversion_) {
+		characterToConversion.emplace_back(character, conversion_);
 	}
 	const char *Find(int character) {
 		const std::vector<int>::iterator it = std::lower_bound(characters.begin(), characters.end(), character);
@@ -690,32 +690,37 @@ public:
 		// Empty the original calculated data completely
 		CharacterToConversion().swap(characterToConversion);
 	}
+	void AddSymmetric(int lower, int upper);
+	void SetupConversions();
 };
 
-CaseConverter caseConvFold;
-CaseConverter caseConvUp;
-CaseConverter caseConvLow;
+CaseConverter caseConvFold(CaseConversion::fold);
+CaseConverter caseConvUp(CaseConversion::upper);
+CaseConverter caseConvLow(CaseConversion::lower);
 
-void AddSymmetric(CaseConversion conversion, int lower,int upper) {
-	char lowerUTF8[UTF8MaxBytes+1];
-	UTF8FromUTF32Character(lower, lowerUTF8);
-	char upperUTF8[UTF8MaxBytes+1];
-	UTF8FromUTF32Character(upper, upperUTF8);
+void CaseConverter::AddSymmetric(int lower, int upper) {
+	const int character = (conversion == CaseConversion::upper) ? lower : upper;
+	const int source = (conversion == CaseConversion::upper) ? upper : lower;
+	char converted[maxConversionLength+1]{};
+	UTF8FromUTF32Character(source, converted);
+	Add(character, converted);
+}
 
-	switch (conversion) {
-	case CaseConversion::fold:
-		caseConvFold.Add(upper, lowerUTF8);
-		break;
-	case CaseConversion::upper:
-		caseConvUp.Add(lower, upperUTF8);
-		break;
-	case CaseConversion::lower:
-		caseConvLow.Add(upper, lowerUTF8);
-		break;
+// Return the next '|' separated field and remove from view.
+std::string_view NextField(std::string_view &view) {
+	const size_t separatorPosition = view.find_first_of('|');
+	const std::string_view field = view.substr(0, separatorPosition);
+	if (separatorPosition == std::string_view::npos) {
+		// Reached the end so empty the view
+		view.remove_prefix(view.length());
+	} else {
+		// Remove the '|' from the view as well as the field
+		view.remove_prefix(separatorPosition + 1);
 	}
+	return field;
 }
 
-void SetupConversions(CaseConversion conversion) {
+void CaseConverter::SetupConversions() {
 	// First initialize for the symmetric ranges
 	for (size_t i=0; i<std::size(symmetricCaseConversionRanges);) {
 		const int lower = symmetricCaseConversionRanges[i++];
@@ -723,91 +728,63 @@ void SetupConversions(CaseConversion conversion) {
 		const int length = symmetricCaseConversionRanges[i++];
 		const int pitch = symmetricCaseConversionRanges[i++];
 		for (int j=0; j<length*pitch; j+=pitch) {
-			AddSymmetric(conversion, lower+j, upper+j);
+			AddSymmetric(lower+j, upper+j);
 		}
 	}
 	// Add the symmetric singletons
 	for (size_t i=0; i<std::size(symmetricCaseConversions);) {
 		const int lower = symmetricCaseConversions[i++];
 		const int upper = symmetricCaseConversions[i++];
-		AddSymmetric(conversion, lower, upper);
+		AddSymmetric(lower, upper);
 	}
 	// Add the complex cases
-	const char *sComplex = complexCaseConversions;
-	while (*sComplex) {
-		// Longest ligature is 3 character so 5 for safety
-		constexpr size_t lenUTF8 = 5*UTF8MaxBytes+1;
-		unsigned char originUTF8[lenUTF8]{};
-		char foldedUTF8[lenUTF8]{};
-		char lowerUTF8[lenUTF8]{};
-		char upperUTF8[lenUTF8]{};
-		size_t i = 0;
-		while (*sComplex && *sComplex != '|') {
-			originUTF8[i++] = *sComplex;
-			sComplex++;
-		}
-		sComplex++;
-		originUTF8[i] = 0;
-		i = 0;
-		while (*sComplex && *sComplex != '|') {
-			foldedUTF8[i++] = *sComplex;
-			sComplex++;
-		}
-		sComplex++;
-		foldedUTF8[i] = 0;
-		i = 0;
-		while (*sComplex && *sComplex != '|') {
-			upperUTF8[i++] = *sComplex;
-			sComplex++;
-		}
-		sComplex++;
-		upperUTF8[i] = 0;
-		i = 0;
-		while (*sComplex && *sComplex != '|') {
-			lowerUTF8[i++] = *sComplex;
-			sComplex++;
-		}
-		sComplex++;
-		lowerUTF8[i] = 0;
-
-		const int character = UnicodeFromUTF8(originUTF8);
-
-		if (conversion == CaseConversion::fold && foldedUTF8[0]) {
-			caseConvFold.Add(character, foldedUTF8);
-		}
-
-		if (conversion == CaseConversion::upper && upperUTF8[0]) {
-			caseConvUp.Add(character, upperUTF8);
+	std::string_view sComplex = complexCaseConversions;
+	while (!sComplex.empty()) {
+		const std::string_view originUTF8 = NextField(sComplex);
+		const std::string_view foldedUTF8 = NextField(sComplex);
+		const std::string_view upperUTF8 = NextField(sComplex);
+		const std::string_view lowerUTF8 = NextField(sComplex);
+
+		std::string_view converted;
+		switch (conversion) {
+		case CaseConversion::fold:
+			converted = foldedUTF8;
+			break;
+		case CaseConversion::upper:
+			converted = upperUTF8;
+			break;
+		case CaseConversion::lower:
+		default:
+			converted = lowerUTF8;
+			break;
 		}
-
-		if (conversion == CaseConversion::lower && lowerUTF8[0]) {
-			caseConvLow.Add(character, lowerUTF8);
+		if (!converted.empty()) {
+			const int character = UnicodeFromUTF8(reinterpret_cast<const unsigned char *>(originUTF8.data()));
+			Add(character, converted);
 		}
 	}
 
+	FinishedAdding();
+}
+
+CaseConverter *ConverterForConversion(CaseConversion conversion) {
+	CaseConverter *pCaseConv = &caseConvFold;
 	switch (conversion) {
 	case CaseConversion::fold:
-		caseConvFold.FinishedAdding();
+		pCaseConv = &caseConvFold;
 		break;
 	case CaseConversion::upper:
-		caseConvUp.FinishedAdding();
+		pCaseConv = &caseConvUp;
 		break;
 	case CaseConversion::lower:
-		caseConvLow.FinishedAdding();
+	default:
+		pCaseConv = &caseConvLow;
 		break;
 	}
-}
-
-CaseConverter *ConverterForConversion(CaseConversion conversion) noexcept {
-	switch (conversion) {
-	case CaseConversion::fold:
-		return &caseConvFold;
-	case CaseConversion::upper:
-		return &caseConvUp;
-	case CaseConversion::lower:
-		return &caseConvLow;
+	if (!pCaseConv->Initialised()) {
+		pCaseConv->SetupConversions();
 	}
-	return nullptr;
+	return pCaseConv;
 }
 
 }
@@ -815,23 +792,16 @@ CaseConverter *ConverterForConversion(CaseConversion conversion) noexcept {
 namespace Scintilla::Internal {
 
 ICaseConverter *ConverterFor(CaseConversion conversion) {
-	CaseConverter *pCaseConv = ConverterForConversion(conversion);
-	if (!pCaseConv->Initialised())
-		SetupConversions(conversion);
-	return pCaseConv;
+	return ConverterForConversion(conversion);
 }
 
 const char *CaseConvert(int character, CaseConversion conversion) {
 	CaseConverter *pCaseConv = ConverterForConversion(conversion);
-	if (!pCaseConv->Initialised())
-		SetupConversions(conversion);
 	return pCaseConv->Find(character);
 }
 
 size_t CaseConvertString(char *converted, size_t sizeConverted, const char *mixed, size_t lenMixed, CaseConversion conversion) {
 	CaseConverter *pCaseConv = ConverterForConversion(conversion);
-	if (!pCaseConv->Initialised())
-		SetupConversions(conversion);
 	return pCaseConv->CaseConvertString(converted, sizeConverted, mixed, lenMixed);
 }